1 Introduction

The seismicity of the Ibero-Maghrebian region (Iberian Peninsula, Morocco, Algeria, and Tunisia) is characterized by the occurrence of large and devastating earthquakes (more frequent in Algeria but of smaller magnitude) and larger ones separated by long time intervals in Iberia and Morocco. The recent occurrence of the disastrous earthquake in Morocco on September 8th, 2023 with MW = 6.8 (USGS), is a good example. During the 20th and 21st centuries, in Iberia, earthquakes that have caused significant damage occurred before 1960, so there are no records of acceleration, but only intensities. For example, the Purchil earthquake on April 19th, 1956 (Imax VII-VIII), caused significant damage and casualties (seven deaths and sixty injuries). The two most damaging earthquakes since the second half of the century are the earthquake located about 180 km southwest of Saint Vincent Cape on February 28th, 1969 (MW = 7.8), and the one in Lorca (southeast Spain) on May 11th, 2011 (MW = 5.1). These two earthquakes generated great economic losses in Spain and caused 4 and 9 fatalities, respectively. For the 1969 earthquake, only an analog strong motion record exists in Lisbon, Portugal (Pro et al., 2020), and for 2011, the PGA values did not exceed 0.36 g (Martínez Solares et al., 2012) at Lorca city (epicentral distance 5 km).

Thus, in any seismic hazard study conducted in Spain, it is crucial to use the macroseismic intensities. These macroseismic intensities must be evaluated with EMS-98 (Grünthal, 1998). The earthquake catalog of the Instituto Geográfico Nacional (IGN online catalog, see Data and Resources) is the database used in seismic hazard studies. Martínez Solares & Mezcua (2002) reviewed earthquake intensities in Iberia up to 1900, standardizing the values using the EMS-98 scale, but there is a lack of review from 1900 to the present date. Although by default, the intensity listed on the website (IGN online catalog, see Data and Resources), regardless of the year, is based on the EMS-98 scale, this is not accurate for earthquakes that occurred between 1900 and 1999. In 2021, the revision and standardization of the focal parameters of earthquakes that occurred from 1900 up to 1962 (the year when the WWSSN stations were installed in Spain) was initiated. In this work, the results obtained for the southeast of Spain are presented since it is the area with the highest seismicity rate during that period. Vannucci et al., (2021) demonstrate a similar issue in the Italian macroseismic database (DBMI15), where the use of intensity scales (EMS-98) in the last decades and MCS (Mercalli–Cancani–Sieberg scale; Cancani, 1904) previously can lead to the underestimation of intensity values. The solutions they propose are a complete reevaluation from the original sources of all intensities in DBMI15, using the EMS. This is exactly the perspective adopted in this work for the IGN seismic database.

Spain is located on the Eurasian–African plate boundary, and most of the earthquakes that occur in this area are due to the convergence between these two plates. Generally, seismicity in Spain is characterized by moderate activity, with large earthquakes separated by long time intervals and shocks occurring at shallow (h < 40 km), intermediate (40 < h < 150 km), and very deep depths (h ≈ 650 km) (Buforn et al., 2004). Figure 1a represents the seismicity for the period 1900–2020 (IGN online catalog, see Data and Resources) with Imax equal to or larger than degree V. When earthquakes have magnitude (ML ≥ 2.9), we represent their epicenters as red circles; in cases where only seismic intensity data is available, we use white circles. This region is called the Ibero-Maghrebian region and lies among Saint Vincent Cape (SVC Fig. 1a), the Alboran Sea (ALB Fig. 1a), southern Iberia, and northern Algeria. In Fig. 1a, we can further observe that other seismic areas are the northern border of Spain, which presents shallow seismicity in the Pyrenees Mountain region (PY Fig. 1a), and the NW area of Spain, which presents moderate seismic activity with some moderate earthquakes from time to time. The northern part of Africa is another seismically active region, in which the greatest activity corresponds to Algeria and the Alboran Sea. The figure also indicates that the main seismic activity in Spain occurs in the S–SE region, where earthquakes are abundant, distributed, and of a greater magnitude than in the rest of the country. This is followed by a NE–SW direction parallel to the coast with an important concentration of epicenters in the western part of the Granada region (GR Fig. 1a).

Figure 1
figure 1

a Distribution of epicenters for the period 1900–2020 (see Data and Resources) for earthquakes with Imax ≥ V. Red circles correspond to earthquakes with magnitudes (ML ≥ 2.9), and white circles are Imax of earthquakes with no magnitude (Imax ≥ V) (GR: Granada region, PY: Pyrenees region, SVC: Saint Vincent Cape region, ALB: Alboran Sea region). b Seismicity (1400–1900), Imax ≥ V; stars correspond to the largest earthquakes

Full size image

During the twentieth century, the highest intensity shocks (Imax ≥ IX) or highest magnitude earthquakes (ML > 6.0) occurred in the Ibero-Maghrebian region, Portugal and northern Africa (Table S1, Supplementary Material).

However, the seismic activity between 1400–1900 was stronger, than in 20th and 21st centuries. Large and damaging earthquakes have occurred in the past in the S-SE of Spain (Fig. 1b), such as the Torrevieja-Alicante earthquake of 1829 (Muñoz et al., 1983) and the Arenas del Rey-Granada earthquake of 1884 (Muñoz & Udías, 1980), both with Imax = IX-X. Other damaging earthquakes are the Montesa-Valencia (1748, Imax = IX) (Buforn et al., 2015) and the earthquakes of Imax VIII-IX that took place in the Granada and Almería regions during this period (Martínez Solares, 1995): Granada (1431), Vera-Almería (1518), Alhama de Almería (1522), Baza-Granada (1531) (Udías et al., 2020) and Dalías-Almería (1804).

Apart from the reevaluation carried out by Martínez Solares & Mezcua (2002) until 1900, other events were re-evaluated later with new information sources, such as Malaga 1497 and 1680 (Goded et al., 2008), Alcoy 1620 and 1644 (Buforn & Udías, 2022), Montesa 1748, Saint Vincent Cape 1755 (Martínez Solares & López Arroyo, 2004), Torrevieja 1829, and Arenas del Rey 1884. In contrast, no systematic re-evaluation has been conducted for the period 1900–1962. Therefore, the study of the 1900–1962 period plays a key role in the homogenization of the seismic database. As an example, different scales were used during this period, and to use the Spanish seismic code (NCSE-02, Norma de la Construcción Sismorresistente Española, 2002), seismic intensities must be given in the EMS-98 scale. It is also important to note that the 20th century is a period in which macroseismic data coexist with instrumental data, resulting in a very heterogeneous set of information. The limit of 1962 corresponds to the time periods outlined in the 'Update of Seismic Hazard Maps of Spain 2012' (Martínez Solares et al., 2013) and pretends to maintain consistency with the previous established standards. These periods have homogeneous information based on the available seismic instrumentation in those groups of years. For the first period, which covers 1900 to 1923, information was mainly gathered from descriptions of earthquake effects and some observations of seismic arrival times. From 1924 to 1962, instrumental information began to be more systematic and was complemented with macroseismic descriptions. Starting in 1962 with the installation of seismic stations of the World-Wide Standard Seismographic Network (WWSSN), instrumental data began to become larger and more predominant.

The main goal of this work is to use a systematic methodology to study the earthquakes that have occurred in Spain during the 1900–1962 period by using information sources that are highly variable in typology and that change throughout the investigated periods. We have used current techniques and scales to re-evaluate and update the seismic parameters of earthquakes. The seismic revision methodology used in this paper was tested in the SE area of Spain to further extend it to the rest of Iberia. Additionally, significant features of this work are the use of traceable data and information sources and the use of reproducible and repeatable calculations.

SE Spain is the most suitable region to investigate and develop a methodology for the study and review of large earthquakes in Iberia. As part of this study, we selected the region between 4°W and 1°E and between 36°N and 39.5°N (Fig. 2) in SE Spain. In fact, this particular area has been affected by a large number of earthquakes over time. Figure 2 represents the seismicity between 1900 and 1962 in the SE region of Spain (IGN online catalog, see Data and Resources) for magnitudes equal to or larger than 2.3 (red circles) or Imax larger than degree II (white circles). The largest earthquakes in this period were Imax VII-VIII earthquakes, which occurred in 1911 at Las Torres de Cotillas, in 1919 at San Miguel de Salinas and in 1956 at Purchil. In general, however, the magnitudes are less than 5.3 MW, and the Imax are less than VIII. The completeness of the IGN earthquake catalog for the studied area (SE Spain) has been achieved for magnitudes MW ≥ 4.5 starting from the 1880s (Martínez Solares et al., 2013).

Figure 2
figure 2

Seismicity for the studied region (period 1900–1962), the red circles events feature magnitudes greater than 2.3 and the white circle events feature maximum Imax greater than II (see Data and Resources)

Full size image

In the first step of our study, we searched for contemporary documents that contained specific information regarding the damage caused by shocks. Then, using this information, a re-evaluation of the focal parameters was carried out to study and compare our results with early solutions.

2 Methodology

Re-evaluating seismic parameters (macroseismic epicenter and EMS-98 intensities) of any earthquake begins by collecting and organizing all available information so that the most appropriate methodology can be chosen. For this purpose, the seismic parameters of the IGN catalog (IGN online catalog, see Data and Resources) and all the original information sources, contemporary documents, and available references have been collected. Then, for each event, we developed an analysis of all the collected information. The following scheme has been used to jointly classify and analyze all the compiled information and resources:

  • All the essential data and basic information required for a subsequent accurate analysis are gathered.

  • A re-evaluation of seismic intensities using the EMS-98 scale is obtained by studying the damage and analyzing the distribution of seismic intensities and the maximum intensity for each event.

  • Re-evaluation of epicenters is performed obtaining instrumental epicenters, macroseismic epicenters, or both, depending on the available information.

  • Conclusions and discussion of results.

In the following subsections, we describe the methodology used to perform the above steps.

2.1 Intensity Evaluation

Macroseismic intensities, and the spatial distribution of their values, are important seismic parameters used to reconstruct the size of an earthquake. It is based on the assessment of the documented effects and damage caused by an earthquake. Different intensity scales have been used throughout the twentieth century. The most commonly used ones in Spain were the ten degrees scale of F-M (Forel–Mercalli scale; Sánchez Navarro Neumann, 1921) and the twelve degrees MSK scale (Medvedev-Sponheuer-Karnik scale; Medvedev et al., 1962). The F-M intensity scale is similar to the MCS, and it is divided into ten degrees that range from instrumental shake to very catastrophic. MSK is a twelve-degree scale but introduces the concept of types of structures, the definition of quantity and the classification of damage to buildings, where degree I is not noticeable, and degree XII involves landscape changes.

The intensity scales used during these years differ in their definitions and in the observation parameters (such as damage in buildings, effects on furniture, people and ground), so it is very important to define a common scale for all the territories and to use it throughout the different time periods. For this purpose, the European Macroseismic Scale (EMS-98) was introduced. This macroseismic intensity scale is well defined, standard, updated and extensively used. It is a twelve-degree scale that evaluates three main features: a) effects on people, b) effects on objects and c) damage to buildings. As a first step, it is necessary to establish the degree of damage to buildings, and for that purpose, it is necessary to evaluate the vulnerability of buildings. Vulnerability ranges from A, which is very vulnerable, such as fieldstone or adobe structures, to F, which is very resistant and includes structures with earthquake-resistant designs. The classification of damage ranges from Grade 1 with no structural damage and slight nonstructural damage to Grade 5 with very heavy structural damage and total or near-total collapses.

In this study, as EMS-98 recommends, if the macroseismic information is not sufficient to assign a single intensity value because there are seismic descriptions that indicate that an intensity could have two different values, then the intensity is expressed as a range of values, such as intensities VI-VII. Furthermore, if the only information available is that the earthquake has been felt or caused damage, but there is not enough information to assign an intensity degree, the intensity can be indicated with an F (felt) or D (damage).

Once all macroseismic information has been reviewed and all re-evaluated intensity data points (IDPs) are available, intensity maps can be drawn. These maps contain all the reviewed intensity data points together. In this case, all the updated intensities can be compared together to study the effects in different places or study local effects. The re-evaluated intensity maps of this work are shown in the Supplementary Material.

2.2 Macroseismic Epicenter

A macroseismic epicenter is established based on the effects caused by the earthquake on people and objects and damage to buildings and structures. If there are enough macroseismic intensity points, mathematical calculation methods can be used to establish the macroseismic epicenter. For this purpose, it is important to have a good azimuthal coverage of the IDPs to surround the epicenter. Also, the more IDPs are available, the more accurate and realistic the obtained macroseismic epicenter becomes. Hence, in cases with a small number of IDPs, mathematical methods have been avoided and the epicenters have been established using the geographical coordinates of the places that suffer the biggest damages and, therefore, experience maximum intensity. This type of epicenter is common in the historical period, before seismic networks were established or at the beginning of their development, when there are no instrumental data or the data are scarce or contain substantial errors. However, these epicenters have low accuracy, and they cannot be directly compared to those determined using instrumental observations (Udías et al., 2020). The mathematical methods used in this work were the Bakun localization method (Bakun & Wentworth, 1997), Boxer 4.0 method (Gasperini et al., 1999, 2010) and MEEP 2.0 method (Musson & Jiménez, 2008). All these methods require a set of IDPs for each earthquake and the locations of those points. A brief description of these three methods of macroseismic location is provided below.

  • Bakun method: This method uses a rectangular grid of coordinates and calculates magnitude MW and RMS for each point of the grid. It locates the event at the point that minimizes the errors of the given IDPs.

  • Boxer 4.0 method: Under Method 0, the epicenter location is determined as the barycenter of the points with maximum intensity. Alternatively, this code uses intensity attenuation models to calculate the hypocenter. Method 1 (B4M1) employs a fixed depth, while Method 2 (B4M2) calculates both epicenter and depth. However, it is worth noting that the depth calculated within B4M2 is usually underestimated by a factor of 2 or more (Gasperini et al., 2010). Boxer 4.0 computes the uncertainty for the epicentral coordinates as the standard deviation divided by the square root of the number of coordinates used.

  • The Macroseismic Estimation of Earthquake Parameters (MEEP 2.0) method includes a calculation with the Meep method, Centroid method, Bakun implementation and a simplification of the Shumila method or Pairwise method. The uncertainty is computed using a bootstrap resampling routine (Musson & Jiménez, 2008).

    1. o

      The Meep method uses a regular square grid in which it considers the IDPs to vary with distance and tries to minimize the residuals (between Imax and the IDPs).

    2. o

      The Centroid method (similar to the Boxer method) starts calculating the centroid with larger intensity values and then continues adding more IDPs until reaching a suitable number of points.

    3. o

      The Pairwise method considers that for two different IDPs, where one is higher than the other, the epicenter should be closer to the largest of them. Then, all possible pairs of distinct IDPs are tested. Therefore, the more pairs of values there are, the more likely it is that the test point will be the epicenter.

2.3 Instrumental Epicenter

The epicenter is calculated using the arrival times of the different seismic phases registered by geophysical observatories. In this paper, the instrumental method used is Hypocenter (Lienert & Haskov, 1995). The inputs of this software are the arrival times of P and S seismic phases from different stations. This method has the capability of locating both regional earthquakes, with a layer velocity model, and teleseismic earthquakes. The same regional model employed by the IGN for location was used in this study, which consists of a 1-D velocity model with flat layers and a Vp/Vs ratio of 1.75 (Mezcua & Martínez Solares, 1983). It is also possible to try an initial hypocenter where it is possible to fix the epicenter and/or depth, or none of them. In addition, it is possible to give weights to the contributions of the seismic phases in the final result, manually given by the analyst, to find the solution with the lowest RMS, by successive iterations. The horizontal (eh) errors are \({e}_{h}=\sqrt{{C}_{xx}^{2}+{C}_{yy}^{2}}\), where C is the elements of the transformed covariance matrix. Errors obtained within this methodology are mainly due to three reasons:

  • Lack of data: A small number of seismic stations were available (Fig. 3), so the data were insufficient to obtain a good quality result.

  • The azimuthal coverage: was poor, due to the network geometry. There were only seismic stations on the Spanish land and most of the epicenters were situated near to the Spanish coast.

  • Accuracy of data: The arrival times from the different seismic phases had to be measured directly from analogical records, and this introduces inaccuracies. In addition, the stations did not have real-time centralized records, and they were not time coordinated, resulting in differences in time systems between stations.

Figure 3
figure 3

Spanish seismic stations before 1962. ROA: Royal Observatory of the Army in San Fernando (Cádiz). JHS: Jesuits Society. ACAB: Sciences and Arts Academy of Barcelona (Academia de Ciencias y Artes de Barcelona). IGN: National Geographic Institute of Spain

Full size image

Therefore, the results obtained by instrumental techniques during the first part of the 20th century can feature large errors and must be analyzed carefully.

When multiple possible epicenters exist (e.g., macroseismic and instrumental epicenters, and re-evaluated intensities in this paper, or epicenters, maximum intensities, and isoseismal maps in other papers), the criterion for selection is the epicenter with the lowest error, located closest to the maximum re-evaluated intensities and damages.

3 Data and Observations

Available seismic information and information sources are different from one earthquake to another, and the types of sources also change between periods. Most of the data used correspond to macroseismic information on the damage caused by earthquakes. There were only 9 seismic stations in Spain during the studied periods (Fig. 3, Table 1), so instrumental data are limited. Seismic stations were first established in Spain by the Royal Navy Observatory in San Fernando, Cádiz (SFS, 1898). However, the first development of Geophysics Observatories with seismological instrumentation in Spain started at the beginning of the 20th century. The first seismic station observatories were established in Cartuja-Granada (CRT, 1902) and Ebro-Tarragona (EBR, 1904), which belonged to the Jesuits Society, and Fabra-Barcelona (FBR,1906), which belonged to the Sciences and Arts Academy of Barcelona. Then, the National Geographic Institute of Spain (IGN) installed seismic stations in the observatories of Toledo (TOL, 1909), Almería (ALM, 1911), Málaga (MAL, 1913) and Alicante (ALI, 1914). In the first decade of the 20th century, observatories were mainly equipped with Vicentini and Bosch Omori seismographs (Batlló, 2004).

Table 1 Spanish seismic stations before 1962
Full size table

Between 1936 and 1939 during the Civil War in Spain and some years later World War II in Europe, seismograms were very incomplete, and a large amount of the information disappeared. From 1950, information from the observatories began to be available again. In Fig. 3, we observe that no seismic observatories were established between 1914 and 1959, when a seismic station was installed in Logroño (LGR, 1959).

Available sources for historic earthquakes can be classified as follows: (a) macroseismic questionnaires, (b) seismic bulletins, (c) seismic catalogs, (e) isoseismal maps, (f) seismograms, (g) newspapers, (h) pictures (and video records), and (i) others. A summary of the studied documents can be consulted in Table S2. There is a description below of the different sources of information used in this study:

3.1 Macroseismic Questionnaires

One of the most important information sources that it is currently possible to find about some earthquakes is represented by macroseismic questionnaires. These documents appeared for the first time in the 1920s (Martín, 1979) and were prepared by seismologists from the IGN. In the beginning, macroseismic surveys were paper application forms with several questions, but only one (number 7 in Fig. 4a) concerning the effects on humans, objects, and damage to buildings:

  • Was it felt by some, many or all the neighbors?

  • Was there any creaking, knocking or falling objects? Which ones?

  • Was there any damage or destruction of buildings?

  • Were there any cracks in the floor? Were there any rockfalls? Was there any influence on the fountains, wells or watercourses?

Figure 4
figure 4

Examples of IGN macroseismic questionnaires: a For the 12/06/1926 earthquake, b For the 03/09/1930 earthquake

Full size image

The IGN sent these forms to the different towns where the earthquake had been felt, and some authority (such as a teacher, personnel from the city hall, a priest, etc.) completed the survey and returned it to the observatories. The collected information from these surveys was then used to evaluate the macroseismic intensity of the earthquake at every observation point, and they were also used to evaluate the Imax of the event.

From the 1930s-40s, macroseismic questionnaires extended the previous survey, including more questions about earthquake effects (Fig. 4b) and asking about damage and effects focusing on different parameters with easy Yes/No questions such as:

  • Felt by someone: Yes/No

  • Felt by few: Yes/No

  • Felt by the majority: Yes/No

  • Felt by almost everyone: Yes/No

  • Felt by all: Yes/No

  • Detachment of plaster from ceilings and walls: Yes/No

  • Small cracks in walls: Yes/No

  • Fall of cornices and tops of towers, etc.: Yes/No

  • Cracks in poorly constructed buildings: Yes/No

  • Cracks in well-constructed buildings: Yes/No

  • Etc.

3.2 Seismic Bulletins

Seismic bulletins came from the observatories of ALI, ALM CRT, EBR, FBR, MAL, SFS, and TOL. They are collections of seismic data, mainly the arrival times of different seismic waves at the seismic station. In addition, these documents can contain more information, such as macroseismic information or damage to the towns around the observatories, or specific information from the observatories, such as the condition of the seismic instruments. For example, Fig. 5a and Fig. 5b show the seismic bulletin from the CRT observatory from July, August, and September, 1945, in which the collected information is the date, time, amplitude, and period of the seismic waves registered by the observatory. This information is completed with additional information on the location of the observatory and preliminary seismic parameters of registered earthquakes.

Figure 5
figure 5

Example of a Seismic Bulletin from CRT (July, August and September, 1945): a Cover and b records of the arrival times of different seismic phases

Full size image

3.3 Seismic Catalogs

Other sources of information are seismic catalogs. The information included in this kind of document includes the seismic parameters of a group of earthquakes, which are elaborated with uniform standards and tested methodologies to provide good quality, complete, and homogeneous seismic data. Usually, the information included in seismic catalogs contains the date of the earthquake, origin time, hypocenter, Imax, magnitude, location, and additional information such as epicenter errors, quality of results, origin of information sources, and macroseismic information.

The seismic catalogs used in this work are Galbis Rodríguez (1932), Galbis Rodríguez (1940), Munuera (1963), Mezcua & Martínez Solares (1983), Vidal Sánchez (1986), Espinar Moreno (1994), Rodríguez de la Torre (1995), and the IGN online earthquake catalog (see Data and Resources).

3.4 Specific Studies

Occasionally, specific studies are conducted for the most important or larger earthquakes. Some focus on analyzing the hypocenter, others examine the effects of the earthquake, others analyze the seismograms, and others look at the magnitude. Figure 6a and Fig. 6b show an example, the study of the Salinas earthquake of November 16th, 1916 (Fontseré, 1917). This study describes the effects of the earthquake in twelve villages (such as Villena and Yecla) together with the P arrival times at 8 stations (Fig. 6b) and an isoseismal map.

Figure 6
figure 6

Example of a specific study (Fontseré, 1917) a Cover page, b Macroseismic information and arrival time of the P wave

Full size image

3.5 Isoseismal Maps

Isoseismal maps are also available for larger earthquakes. This information source must be used with care because in this kind of map, two datasets are represented: intensity points and isoseismal lines (which are isolines that separate regions of different intensities). The problem is that the traces of the isoseismal lines simplify the original data and, in some sections, are not accurate enough or even wrong. Furthermore, to draft these maps, sometimes some IDPs are supposed to support the isoseismal lines, but currently, it is not possible to distinguish the two kinds of points. An example of an isoseismal map is shown in Fig. 7 (Rey Pastor, 1932). In this map, it is possible to distinguish the isoseismal lines and the IDPs, but it is not possible to know the original information used to plot this map.

Figure 7
figure 7

Isoseismal map from the 05/03/1932 earthquake, Rey Pastor (1932)

Full size image

3.6 Seismograms

Seismograms are a very useful information source because they represent the original source of the arrival times of different seismic waves that make it possible to study the instrumental location of earthquakes. In a few cases, such data are available. However, sometimes the seismograms are mere photographic reproductions or have been obtained from a report. In Fig. 8, we show a seismogram from EBR (E–W component) for the September 10th, 1919, earthquake.

Figure 8
figure 8

Example of a seismogram: The E–W component record of the 10/09/1919 earthquake by the Vicentini Seismometer at the EBR station

Full size image

3.7 Newspapers

Larger earthquakes also have an important impact on society, so local or even national newspapers usually describe their effects. This type of information is not as rigorous as a scientific report, and it may not be very accurate or even exaggerated; therefore, it must be evaluated with care.

For instance, by evaluating the information of the damages caused by the Lúcar earthquake March 5th, 1932, in a local newspaper (Crónica Meridional, Almería, 9/3/1932), the injuries suffered by Juan Trujillo Martos due to the earthquake are described as 'hurting in the left leg'. However, these damages are heightened in a national newspaper (La Voz, Madrid, 10/03/1932), where the injuries have worsened, and the condition of the same individual is described as 'being on the brink of death'.

However, collecting as much information as possible from newspapers makes it possible to obtain information about the effects of the earthquake on different buildings, towns, etc., which can be very useful for the study.

3.8 Pictures (and Video Records)

Since the beginning of the 20th century, pictures of earthquake-related effects have occasionally been found, although not very often. It is an original and primary source that allows us to evaluate the damage with no previous interpretations. Pictures can be found in scientific reports and newspapers or were sometimes taken for a particular study of an earthquake (Fig. 9). From these pictures, it is possible to evaluate the degree and type of damage, a building’s vulnerability, and the seismic intensity in a particular place. Although video records are very scarce, we have found some in the Spanish television program NODO (NOticiario y DOcumentales, News and Documentaries, see Data and Resources) with information about earthquake-induced damage.

Figure 9
figure 9

Example of a picture: Damage caused by the 10/09/1919 earthquake in Torremendo town

Full size image

3.9 Others

  • Local archives of churches and city halls: The complementary information obtained from the archives of local churches and city halls is very useful to extend and complete the data obtained from the abovementioned information sources.

  • Seismic phase records: Finally, in some observatories, in addition to seismic bulletins, there are sometimes collections of seismic phase records directly obtained from seismograms. This is a preliminary product that precedes the construction of the seismic bulletins, but unfortunately, these records are not always available.

3.10 Selected Earthquakes

For this study, 16 earthquakes have been selected from the IGN earthquake catalog (Table 2, Fig. 10a and Fig. 10b). These earthquakes occurred in the period 1900–1962 in the study area, between 4°W and 1°E and between 36°N and 39.5°N. Fourteen events have Imax equal to or larger than VII. Additionally, two earthquakes with lower Imax (IV and VI) have been included in the analysis based on newly discovered data.

Table 2 Selected earthquakes from the 1900–1962 period (IGN online catalog, see Data and Resources)
Full size table
Figure 10
figure 10

Selected earthquakes that occurred in the studied region in the a 1900–1923 period and b 1924–1962 period

Full size image

For the selected earthquakes, three earthquakes have Imax = VII-VIII, eleven have an Imax = VII, another has an Imax = VI, and the remaining earthquake (April 25th, 1912) has an Imax = IV (Fig. 10a and Fig. 10b). We can also observe that most of the selected earthquakes are located in the area where large earthquakes occurred in 1748 (Imax = IX), 1829 (IX-X) and 1884 (IX-X) (Fig. 1b), that was an important criterion for stablish the area of study.

Given that three of the largest shocks, which occurred in March and April of 1911 and September of 1919, were previously examined by various authors, Buforn et al. (2005) and Batlló et al. (2015), employing modern methodologies, we have relied on their results. In this work, we used all the available contemporary information to re-evaluate the seismic intensities, macroseismic epicenters, instrumental epicenters, and, for eight earthquakes, the depth.

4 Results

In Table 3, the results of the reevaluation of intensity and hypocenter are shown. The Supplementary Material (Tables S4 to S37) includes all the data used for the reevaluation of intensity.

Table 3 Re-evaluated seismic parameters for the 1900–1962 period in this study
Full size table

The main differences between our study and the IGN earthquake catalog correspond to the assigned maximum intensity (with variations greater than one intensity degree for events EQ5, EQ8, EQ12 and EQ13) and in the epicenter locations (with variations greater than 10 km for events EQ7, EQ10, EQ11 and EQ15) (Table 3, Fig. 11a and Fig. 11b). The rest of the events, EQ1, EQ2, EQ3, EQ4, EQ6, EQ9, EQ14, and EQ16, have minor changes or no changes in Imax or the epicenter. In addition, we have estimated the focal depth for events (EQ2, EQ4, EQ10, EQ12, EQ13, EQ14, EQ15, and EQ16), which were not available in the IGN. However, according to Gasperini et al. (2010), we must take into account that these depths may be underestimated. The original Spanish descriptions used for the macroseismic study are in the Supplementary Material (Table S3).

Figure 11
figure 11

Re-evaluated earthquakes (colored circles) versus IGN epicenters (black circles): a 1900–1923 period, b 1924–1962 period. Seismic intensity EMS-98 has been indicated in the figure legend

Full size image

While a mathematical tool is preferable for reducing subjectivity, we have compared the epicenters obtained through mathematical methods with those derived from Macroseismic Information (Table S38) in cases where there is a limited number of available IDPs. The results reveal no significant differences, with only 0 to 2 km variation between them.

However, in earthquakes with only 1 IDP, the error provided by a mathematical method is not considered representative. This is because the epicenter obtained with B4M1 or B4M2 corresponds to the coordinates of the single IDP itself. Moreover, one of the objectives of this work is to update the IGN earthquake catalog for the period 1900–1962 while maintaining its homogeneity. Following the criterion used by Martínez Solares & Mezcua (2002), for earthquakes with a low number of IPDs, we choose to assign an epicenter based on Macroseismic Information (M-I). Therefore, we prefer the M-I epicenter for these earthquakes.

The estimation of an instrumental epicenter has been possible only for EQ11. Although some of the methods used provide an estimation of the magnitude based on intensity, we have chosen not to use that information, as we believe it should be derived from seismograms.

In the following, we detail the eight earthquakes exhibiting the most significant variations in Imax or epicenter in comparison to the IGN catalog.

4.1 Changes in Imax

The four earthquakes with the largest variations in Imax (variations greater than one intensity degree) between the IGN earthquake catalogue and this study, are shown below in chronological order:

4.1.1 EQ5 (22/04/1912)

Earthquake EQ5 (black circle in Fig. 11a) has an assigned maximum intensity of VII (Forel–Mercalli) in the IGN catalog. For this earthquake revision, the only available macroseismic information corresponds to the village of Ocaña (Almería) (Fig. S5), where many buildings exhibited damage (Boletín Mensual de la Estación Sismológica de Cartuja (Granada), 1912), and there is no report or information for any other place.

Descriptions in newspapers of the damage in Ocaña village are as follows: The earthquake has caused much damage to the buildings. Some houses have been destroyed (El Día, April 24th); many houses have suffered damage, and others are ruined (La Independencia, April 24th); some buildings collapsed, and others cracked. The neighbors, frightened, went to the field (La Cruz, April 26th).

The traditional structures in this region were country houses and farmhouses, most of which were built with earthen bricks or dry stone, later whitewashed and without foundations (see Data and Resources). Many examples of this type of construction can still be seen today in this village. Therefore, these buildings probably had a vulnerability of A on the EMS-98 scale and exhibited very poor maintenance conditions. And, as EMS-98 indicates: A building which has been allowed to decay may be significantly weaker, sufficiently to reduce it by at least one vulnerability class. But in this case, it is not possible to assign a vulnerability smaller than A.

In addition, Sánchez Navarro Neumann (1913), using the seismogram recorded at station CRT and comparing it with those of other similar earthquakes in this area, concluded that the effects of this earthquake could have been exaggerated due to the poorly constructed buildings. Moreover, there is no mention of damages or any other effects in the surrounding villages of Doña María de Ocaña (1180 inhabitants versus 910 in Ocaña), Los Lázaros, La Mosca or Los Soleres. These towns, situated in the northeast of Ocaña and located only one to two kilometers from the epicenter, show no reported damage. The damage in Ocaña may be due to the directivity or local site amplification effects.

Thus, the macroseismic intensity (EMS-98) could have been VI or VII with a vulnerability of A in buildings with well-constructed structures. However, in this case, the poor condition of the buildings reduces the most likely maximum intensity value to V or VI (EMS-98).

To summarize, the poor conditions of the buildings in Ocaña and the lack of damages in the surrounding villages (Fig. S5) suggest that the size of this earthquake may have been overestimated, and an Imax of VII may not be justified.

For re-evaluating the epicenter, there is not enough instrumental information available to calculate an instrumental epicenter. Despite the fact that the IGN epicenter is located in Santa María de Ocaña, no macroseismic information is available regarding this particular place.

In conclusion, we suggest updating the seismic Imax to V-VI (EMS-98) and moving the epicenter from IGN’s epicenter to 37.13° N and 2.73° W (coordinates of the village of Ocaña, Almería) (Fig. 11a, dark green circle and Table 3) approximately 1 km from IGN’s epicenter.

4.1.2 EQ8 (25/11/1913)

This earthquake is part of a seismic sequence (IGN seismic catalog) of at least 9 reported felt earthquakes that started on November 24th, 1913, at 22:15 UTC and finished on the 25th at 6:30, with the main earthquake occurring on the 25th at 2:27 UTC.

The macroseismic information of this earthquake (Fig. 11a and Fig. S8) was obtained from newspapers and a scientific report (Sánchez Navarro Neumann, 1914). There are also arrival times of seismic waves recorded at the stations CRT and SFS, but they are not sufficient for instrumental revision.

There are two different types of information from the newspapers: (a) the damage was light or moderate: no misfortunes have occurred. Some buildings have suffered (El Imparcial, November 29th); many neighbors have abandoned their homes (El Globo, November 29th); Twenty seismic shocks have been felt in Huéscar, some quite intense. The persistence of the phenomenon has been a cause of real panic in the town and the surrounding area (Diario de Valencia, November 29th). (b) Other sources say that the earthquake has not caused any damage: […] but that they have had no consequences (El Pueblo, November 29th); Fortunately, they did not cause any other effects than those of the natural alarm among the inhabitants (El Salmantino, November 29th).

Sánchez Navarro Neumann (1914) provided the following information about damage to buildings in Huéscar: Many partition walls have fallen down or cracked, and many chimneys have been ruined, without having to lament, however, any personal misfortune. In Galera, a village 7 km south of the first one also felt the earthquake: they have also noticed [the earthquakes] in the immediate place of Galera (El País, November 29th); this phenomenon has also been noted in the town of Galera (La Correspondencia de España, November 29th).

Considering that the vulnerability of structures in these two towns was of class A, buildings could present damage of degree 1 or 2, so the maximum seismic intensity in Huéscar (EMS-98) would be V or VI. Therefore, the IGN Imax = VII is not justified with the present data.

The IGN epicenter of earthquake EQ8 (black circle in Fig. 11a and blue star Fig. S8) (Mezcua & Martínez Solares, 1983) has been calculated using Hypo71. However, this is probably a low-quality result, as it is based on one single P and S observation. As a result, we prefer a macroseismic epicenter to the S of Huéscar located at 37.80° N, 2.55° W (dark green circle in Fig. 11a), taking into account that the effects were greater in Huéscar but were also felt in Galera. The M-I epicenter is in close proximity to the B4M1 epicenter (1 km apart, see Table S38). The maximum seismic intensity corresponds to V-VI (EMS-98) in the village of Huéscar.

4.1.3 EQ12 (05/03/1932)

Earthquake EQ12 (the IGN epicenter is represented as a black circle in Fig. 11b) has been revised using the available macroseismic information (in this case, newspapers). The largest damage corresponds to Lúcar (Fig. 12): The earthquake […] caused damage to many houses and detachment of stones from the hill called “Los Cámaras”, which throw down two buildings (Crónica Meridional, March 9th); […] causing some important damage, especially on number 5 of Carmen street, on whose roofs a large number of stones from "Las Cámaras" hill collapsed, sinking the roofs of the mentioned house […] Also, although less important, the Chapter House and the Civil Guard barracks suffered damage (Diario de Almería, March 9th). It is very difficult to distinguish the effects of the fallen rocks from the earthquake effects, but it is evident that the earthquake produced many cracks in buildings that were amplified by fallen stones from nearby mountains: The whole town was outside their homes because many of them had cracks (La Independencia, March 9th); The first seismic phenomenon was perceived, lasting approximately thirty seconds and with such intensity that in all the houses there was not any "unbroken crystal" (La Independencia, March 9th); An earthquake was felt causing significant damage (La Voz, March 9th). And in the rest of villages the earthquake was felt but no damage was reported.

Figure 12
figure 12

Results for EQ12 earthquake, the red star (in both panels) corresponds to the assigned epicenter of this study. a Epicenters obtained with the macroseismic methods and the location errors, blue star corresponds to the IGN epicenter and the red lines correspond with the geological faults. b Final result, the circles correspond to the re-evaluated IDPs with EMS-98

Full size image

Therefore, with this information in hand, in buildings where the most likely vulnerability was A or B, damages were of degree 1 or 2. Therefore, the maximum seismic intensity in Lúcar was between V and VI (EMS-98). Collateral damage was caused by the rockfall from nearby mountains. The damages on a roof were produced by stones from the nearby hills, so it is an isolated case observed in a single house (Table S2, Row 2 of EQ12). Our conclusion is that the IGN Imax is not justified, and we propose an Imax (EMS-98) of V-VI versus VII assigned by the IGN (Table 3).

The instrumental epicenter obtained in this study presents errors of 7 km. Additionally, this result is more than 11 km from the largest damage of this earthquake. Hence, we have calculated a macroseismic epicenter using the 46 intensity data points from the isoseismal map (Fig. 7) of Rey Pastor (1932) and by means of B4M1 and B4M2, obtaining the same result with both methods. The macroseismic epicenter obtained (Table 3 and Table S38) is located at 37.40° N, 2.41° W, with an error of 3 km. This location aligns well with the 15 revaluated intensities (EMS-98) presented in this work (Fig. 12). Our epicenter is located approximately 3 km SE of the location given in the IGN seismic catalog.

4.1.4 EQ13 (23/02/1944)

Earthquake EQ13 took place in Fortuna village (indicated by a black circle in Fig. 11b) at 22:34 UTC. This means that everyone was surely asleep or at least at home, even though there is no news or description of this event in Fortuna’s epicentral area. At the time of the event or in the days afterward, there was no news about the effects or damages in Fortuna a village with 6500 inhabitants at that time. The church, palace and castle were not reported as damaged by the earthquake. However, the IGN has assigned a maximum intensity of degree VII.

In Murcia (Fig. S12), the effects of the earthquake were reported in newspapers as follows: quite intense shakes were felt (ABC, February 24th); there was movement of pictures, doors and lamps, and an intense noise that caused alarm in the neighborhood (Patria, February 24th); The shaking was so intense that in some places the furniture moved and the light bulbs that hung from the wires oscillated noticeably (El Adelanto, February 24th). The described effects in Murcia allow us to conclude that a maximum intensity of IV-V (EMS-98) is likely for this place (which is not the epicentral area). The IGN maximum intensity presents a degree VII, what doesn’t seem appropriate due to the lack of documented damage in Fortuna. In addition, due to the loss of the original macroseismic descriptions, it has only been possible to re-evaluate the seismic intensity (EMS-98) in 7 places.

The instrumental epicenter obtained in this study present large errors, so it has only been possible to estimate a macroseismic epicenter using the 73 IDPs obtained from Rey Pastor (1944). We have obtained the epicenter using both B4M1 and B4M2 methods, locating it at 38.17° N, 1.14° W, which is very close to the IGN epicenter, and with a shallow depth (h < 10 km). The uncertainty of the epicenter is 1 km.

4.2 Changes in the Epicenter

The four earthquakes with the largest variations in the epicenter location (greater than 10 km) between the IGN earthquake catalogue and this study are shown below in chronological order:

4.2.1 EQ7 (11/08/1913)

The greatest epicentral location change in our study is for EQ7, 41 km between the IGN epicenter and this study. Some sources, including the IGN seismic catalog, locate this event in the town of Albuñol (black circle in Fig. 11a). However, according to the original information, the correct location is Albuñuelas (orange circle in Fig. 11a). The data that support this change are as follows:

  • Newspapers indicate the following: […] the village of Albuñuelas […] felt a great earthquake accompanied by a formidable noise that lasted several seconds. The neighborhood hastily abandoned their houses. The panic that ensued was indescribable. There have been no personal misfortunes, but the buildings have suffered damage (El Defensor de Granada, August 12th); in Albuñuelas […] There was great trepidation. The entire neighborhood went to the streets, settling out of reach of the buildings, in case they collapsed. The panic was terrible (Gaceta del Sur, August 13th). Therefore, Albuñol is never mentioned, but Albuñuelas was mentioned twice.

  • The newspaper Gaceta del Sur (Granada), from August 13th, describes the observations of the priest of Albuñuelas, Mr. Gámez, during the earthquake: Many neighbors wanted the bells to ring so that no one would stay at home, but the parish priest, Mr. Gámez, did not agree to prevent the alarm from increasing. He did not give his consent for people to gather in the church in case the phenomenon repeated, as was to be feared, there would be outrages and misfortunes. Searching the records of the Albuñol and Albuñuelas churches confirms that Mr. Gámez was the parish priest of the Albuñuelas church and that there is no mention of Mr. Gámez in the Albuñol church achieves.

  • In addition, the seismic bulletin Boletín Mensual de la Estación Sismológica de Cartuja (Granada) (1913) locates a preliminary epicenter approximately 30 km to the SSW of the CRT observatory in Albuñol town. However, in reality, the town 30 km to the SSW is Albuñuelas, not Albuñol, which is approximately 56 km to the SE of the CRT observatory (Fig. S7), in Granada.

Based on these data, we propose for the EQ7 earthquake a macroseismic epicenter located at 36.93° N, 3.63° W in the town of Albuñuelas, approximately 41 km from the IGN epicenter (orange circle in Fig. 11a and Table 3).

We agree with the IGN maximum intensity value of this earthquake, namely, VII (EMS-98), but we also provide new information: a set of 5 IDPs are available (Fig. S7).

4.2.2 EQ10 (10/09/1919)

This earthquake (black circle in Fig. 11a) has been the object of several detailed studies re-evaluating the seismic parameters (Batlló et al., 2015; Garrido, 2012; Martínez Solares, 2014; Martínez Solares & Batlló, 2017). No new information has been found, so we accept the IGN maximum intensity value (VII-VIII EMS-98, Table 2), but we have re-evaluated the epicenter using different data: (a) the arrival times at stations ALI, ALM, CRT, EBR, MAL, TOL, FBR and SFS (Fig. 3 and Table 1); (b) the seismic questionnaires kept in the seismic Data Archive in the IGN headquarters; and (c) the intensity re-evaluation performed by Batlló et al. (2015).

We calculated the epicenter using two methods. The first one is an instrumental epicenter using the Hypocenter method (Lienert & Haskov, 1995), which yielded an offshore epicenter, far from the greatest damage located in Torremendo, Benejúzar, Jacarilla, and Algorfa towns and with a locational uncertainty of 20–23 km (Fig. S10). This uncertainty is due to the network geometry (Fig. 3), characterized by a gap of 224º because the epicenter is near to the Spanish coast and there are only seismic stations on the Spanish land. The second approach involves determining the macroseismic epicenter using B4M1 and B4M2, considering 83 re-evaluated EMS-98 IDPs (Batlló et al., 2015). The result is a macroseismic epicenter located at 38.07° N, 0.82° W (red circle in Fig. 11a and red star in Fig. S10), and a shallow hypocenter (h < 10 km), approximately 11 km to the NE from the IGN epicenter and with uncertainties less than ± 2 km (Table 3). Therefore, we propose this value as a macroseismic epicenter, and we maintain the maximum intensity of the IGN seismic catalog (VII-VIII EMS-98, Table 3).

4.2.3 EQ11 (12/06/1926)

For EQ11, the information available is (a) the arrival times at stations ALI, ALM, CRT, EBR, FBR, MAL, TOL, and SFS; (b) newspapers; and (c) the seismic questionnaires kept in the seismic Data Archive in the IGN headquarters.

Maximum seismic intensities have been revised using the information of the macroseismic questionnaires and completed with newspaper information. We assigned seismic intensity values (EMS-98) in 14 towns, and at 8 sites, the earthquake was felt, but it was not possible to assign an intensity value. The maximum intensity assigned is V (EMS-98) (Table 3 and Fig. S11) in Almeria: The walls of some old or modest houses have suffered the effects of the earthquake (La Independencia, June 13th); Some old buildings cracked (Diario de Burgos, June 14th); Two houses on the Paseo del Príncipe, among other buildings, and some other houses had cracks in their upper part (La Independencia, June 15th). This maximum intensity is lower than the IGN value (VI, Table 2).

The epicenter has been relocated using the Hypocenter method (Lienert & Haskov, 1995) and 9 P and 6 S wave observations. This new instrumental offshore epicenter is located at 36.68° N, 2.43° W (blue circle Fig. 11b), 11 km to the SW of the IGN epicenter (black circle Fig. 11b), but with large uncertainties (semiaxes error ellipse 27 km in latitude and 42 km in longitude). It has not been possible to calculate the depth.

4.2.4 EQ15 (23/06/1948)

For the review of the IGN solution of EQ15 (the black circle in Fig. 11b and the blue star in Fig. S14), the main available information is (a) the seismic waves recorded at the ALI, ALM, CRT, EBR, FBR, MAL, TOL, and LIS stations; (b) an isoseismal map by Rey Pastor (1951); and (c) damage descriptions in newspapers.

We have reviewed the available information, and we confirm VII (EMS-98) as the maximum seismic intensity. In addition, we provide a set of 6 revised IDPs with values that range from IV-V to VII (EMS-98).

The instrumental epicenter (using the Hypocenter method) presents large errors (approximately 20 km), so we have calculated a macroseismic epicenter using B4M1 and B4M2, and 93 IDPs from Rey Pastor (1951). The best result corresponds to a macroseismic epicenter at 38.15° N and 1.76° W (orange circle in Fig. 11b, Table 3) with an error of 4 km. It has a shallow hypocenter (h < 10 km). This macroseismic epicenter is located between the two most damaged towns, Valentín and Cehegín (Fig. S14).

5 Discussion

In the first period (1900–1923), the intensities and epicenters of 10 earthquakes were re-evaluated using original documents and, in some cases, new information. Our re-evaluated seismic intensity (EMS-98 scale) are lower than the IGN Imax for 4 events (EQ2, EQ5, EQ8, and EQ9). This variation ranges from one to one-and-a-half degrees and may be explained by different factors. The first is the vulnerability of structures or the very poor maintenance conditions of buildings that were not considered in previous studies or catalogs (EQ5). Another factor is the intensity scale used (Forel–Mercalli), which is outdated; therefore, we reviewed all the contemporary sources with the EMS-98 scale (EQ2). Experience shows that comparing different macroseismic intensity scales is far from straightforward. Intensity values could vary more between researchers using the same scale than values change from one scale to another if the researcher remains the same (Grünthal, 1998). Only for event EQ6 we conclude that the IGN Imax was underestimated (IV versus IV-V in our study). The discrepancy arises because various newspaper reports describe the earthquake differently: some indicate damage to buildings and houses with cracks (intensity V, EMS-98), while others, such as the description from the priest of Lorquí, mention widespread shaking but no damage to structures (intensity < V). These contrasting reports suggest that newspapers may have presented exaggerated descriptions. With the available information, there are indicators of both intensity IV and V, making it impossible to assign a single value. The most accurate Imax (EMS-98) falls within the range of IV-V.

For the second period (1924–1962), we re-evaluated the intensities and epicenters of 6 earthquakes. For 5 events (EQ11, EQ12, EQ13, EQ14, and EQ16), our Imax is lower than those of the IGN ranging from a half degree (EQ14 and EQ16) to two-and-a-half degrees (EQ13). The largest variations in Imax are due to a lack of damage information in the epicentral area and surroundings (EQ13). Minor changes in Imax are due to the re-evaluation of intensity with the EMS-98 scale and consideration of the vulnerability of buildings (EQ11, EQ12, EQ14, and EQ16).

Regarding the epicenters from the first period, we have updated eight of the ten earthquakes studied. In this period all epicenters are obtained from macroseismic information due to the lack or low number of seismograms to obtain an instrumental epicenter. The largest difference with the IGN catalogue is for EQ7, where the epicenter has been moved 41 km to the NW due to a mistake with the town’s name with largest damage. The correct epicenter is located in the town of Albuñuelas based on information from contemporary newspapers, church achieves, and seismic bulletins (Table 3).

For events EQ2, EQ4, and EQ10, macroseismic epicenters were determined using either B4M1 (EQ2 and EQ10) or the MEEP method (EQ4) (referred to as M-M2C in Table 3). Additionally, the study included an examination of depths, revealing shallow earthquakes (h < 10 km). In the case of the EQ2 earthquake, the same epicenter as the IGN is obtained. The epicenter of EQ4 is located 2 km away from the IGN value with an uncertainty of 3 km. The most significant deviation from the IGN epicenter was observed for EQ10 (11 km difference). In this case as well, the uncertainty for the epicenter is relatively small, being 2 km.

For EQ1, EQ5, EQ6, EQ8, and EQ9 events, our epicenters have changed in comparison to the IGN seismic catalog between 1 and 5 km. These epicenters are obtained using macroseismic information (named M-I in Table 3). For EQ9, despite calculating an instrumental epicenter, it has been excluded from consideration due to a substantial error of approximately 80 km. This large error is attributed to the limited data available (only 8 P arrivals) and the inadequate azimuthal coverage of the network geometry (105º gap). Due to these limitations, our preferred choice was the macroseismic epicenter.

The maximum differences between the B4M1 and M-I epicenters are 2 km (for the earthquakes EQ1 and EQ9), and in both cases, only 2 IDPs are available. Nevertheless, there are cases where the difference between the B4M1 and M-I epicenters is 0 km (for earthquakes EQ5, EQ6, and EQ7), indicating an identical epicenter regardless of the method used (see Table S38).

The epicenters of the second period were re-evaluated by calculating both macroseismic and instrumental epicenters. For EQ11, the instrumental epicenter (named I-HY in Table 3) was assigned because it is the most adequate solution for an offshore earthquake, despite the large errors of the epicenter 27 km and 42 km. For EQ12, EQ13, EQ14, EQ15, and EQ16, we established macroseismic epicenters with B4M1 by re-evaluating the available contemporary information sources (Table 3). The differences of these epicenters with the IGN are of 4 km or less, except for EQ15 (11 km) (Table 3). The uncertainties for the epicenters are minimal, measuring less than 4 km. Additionally, we have determined the focal depth for these five events resulting in shallow foci.

The estimated depths, obtained through B4M2, should be regarded as indicative of the shallow nature of these earthquakes, as they may be underestimated (Gasperini et al., 2010). Assuming a potential underestimation of about 50%, it is reasonable to consider that the actual depths of these earthquakes are likely less than 10 km. Other studies conducted in the same area, using instrumental and broadband recordings, have consistently demonstrated shallow focal depths (h < 10 km) (Mezcua et al., 1984; Buforn et al., 2005, 2006; Batlló et al., 2015). Moreover, the earthquake's impact is confined to a limited area, reinforcing the notion of shallow foci.

6 Conclusions

From this study and with the revision of the 16 earthquakes, we conclude the following:

  • We have used a systematic methodology for the re-evaluation and study of earthquakes in the first decades of the twentieth century in the S-SE of Spain, a period in which the information sources are very heterogeneous.

  • The joint use of all available information, including individual reports or studies, newspapers, pictures, and seismograms, is critical. An essential requirement is that these documents must be original and not obtained from other sources or authors, as the latter have likely been modified or interpreted.

  • For both 1900–1923 and 1924–1962, some of the IGN Imax values were larger (9 events over a total of 16) by between a half degree to two-and-a-half degrees, due to the different scales used. These differences may be explained by the fact that the previous intensity scales do not consider the vulnerability of buildings, whereas this consideration has been taken into account in this study using the EMS-98 scale. Therefore, the different seismic scales cannot be direct compared.

  • The epicenters in this work have been re-evaluated with macroseismic and instrumental information when possible. However, in the two studied periods, most epicenters have been revised based on macroseismic information.

  • In this study, macroseismic epicenters have proven to be the most suitable methodology due to the intrinsic nature of the data, as opposed to instrumental data which were scarce or less precise.

  • In general, these macroseismic epicenters are well determined with uncertainties less than 5 km. For 8 of the earthquakes, it has been possible to estimate the depth, resulting in shallow foci (less than 10 km).

  • This re-evaluation reveals the importance of the macroseismic intensities, that are a key parameter to carry out any seismic study in this period when instrumental information is scarce or non-systematic. In addition, in Spain, there is a limited availability of acceleration records, and the existing ones often exhibit low values. In this context, macroseismic intensities play a crucial role in ensuring an accurate estimation of seismic hazard.

  • We have compilated all the original, contemporary, and singular information sources. All the individual information sources consulted are free access, but this information together-collected will be available in the future at the IGN web site. The revised information will be included in the earthquake catalogue of the IGN. This updated and homogeneous dataset could be used for different purposes, including: (a) revising and updating the seismicity maps of Spain, (b) revising and updating the seismic hazard maps, (c) revising and updating the seismic construction standards, and ((d) studying the seismicity of important sites such as nuclear power stations, dams, gas storages and more.